Control of Precipitation of Cellulose Solutions in N-Methylmorpholine-N-oxide by Introducing Polyacrylonitrile Additives

Abstract

The precipitation of cellulose and polyacrylonitrile and its copolymer (PAN) solutions is a well-known process that has been extensively described in numerous studies. It is suggested that “soft” precipitants (aqueous solutions of solvent, alcohols) be used in place of “rigid” ones (water) to control the rate at which solutions precipitate. Diffusion processes can also be controlled by lowering the temperature of the interacting system’s constituent parts. The appearance and structure of the resulting fibers (films) are directly correlated with the rate of coagulation. Adding a composite additive to the solution is an unusual method of altering the rate of polymer phase release. The introduced additive should dissolve in a common solvent, which will ensure the competition of precipitation between the polymer phases. It is shown that using optical methods it is possible to trace the evolution of the polymer phase precipitation and the formed morphology. For 12% solutions of cellulose, PAN and mixed systems in N-methylmorpholine-N-oxide (NMMO) the kinetics of the movement of isoconcentration planes was traced and the growth rates of the precipitated polymer zone were estimated. The introduction of PAN additives into cellulose enables the influence of diffusion processes and minimizes the formation of finger-like defects (vacuoles). When the PAN content in the system is 30% or more, the formation of defects in the precipitated solution is significantly suppressed, which is crucial for achieving a uniform morphology.

1. Introduction

The most widely used polymers for producing carbon fiber (CF) precursors are cellulose and polyacrylonitrile and its copolymers (PAN) [1,2]. The demand for CF based on the polymers does not decline in spite of their differences. The reason for this is that cellulose-derived CF has a low thermal expansion coefficient [3]. High-strength and modular CF may be obtained by using PAN precursors [4].

The direct solvents of cellulose and PAN are determined by their chemical structures [5,6]. The most often used solvent for cellulose is N-methylmorpholine-N-oxide (NMMO), whereas the principal solvents for PAN include DMSO, DMF, DMAc, and sodium thiocyanate [7,8]. It has recently been demonstrated that PAN solutions with a concentration of more than 50% may be obtained by using NMMO [9]. This has enabled the production of cellulose and PAN-based mixed solutions [10].

The conditions for producing solutions and their development into fibers (films) are determined by the properties of the NMMO solvent [11]. Briefly, the spinning solution with a concentration of 10 to 25% is passed through a multichannel or slotted spinneret, after passing through an air gap, the just formed solution enters into a precipitation bath [12]. Mass exchange processes between the precipitant and the solvent start as soon as they come into contact with the liquid [13]. A liquid that is compatible with the solvent and does not dissolve the polymer is called a precipitant [14,15,16]. Polar solvents such as water, alcohols, aqueous alcohol solutions, and aqueous NMMO solutions are examples of compatible liquids [17,18,19].

As a numerical characteristic of the precipitation process, it is proposed to use the diffusion coefficient, which depends on the nature of the selected precipitant. For the MMO-process, the numerical values of solvent (solution) diffusion into water are described in [20,21,22,23]. The rate of these processes and the release of the polymer phase is associated with the interaction between the precipitant and the solvent [24], the concentration of the solution, temperature and the type of precipitant [25,26]. The precipitant’s diffusion into the solution volume is slowed considerably as its temperature drops [27]. By creating a homogeneous structure, the application of “soft” precipitants, on the other hand, slows down the release of the polymer phase [26,28,29]. A change in the viscosity (concentration) of the solution also affects the precipitation rate and the morphology of the formed material [30,31].

The phase diagrams serve as the basis for modeling the processes of creating spinning solutions and further separation of the polymer phase [32]. It should be noted that while such diagrams exist for cellulose and PAN [33,34,35,36], they are not available for mixed systems.

A detailed description of the polymer phase’s precipitation for cellulose has been provided [22]. Aqueous solutions of NMMO have been shown to be optimal liquids for the precipitation bath [26]. Replacement water with aqueous solutions of NMMO allows not only to achieve the required structure in the formed material, but also simplifies solvent regeneration. Interestingly, aqueous solutions of NMMO are also well suited for the precipitation of PAN solutions [36]. The optimum concentration of NMMO in the precipitant is about 20%. Comparison of precipitation numbers for cellulose and PAN solutions in NMMO revealed that a smaller amount of water is required for the latter’s coagulation. In other words, PAN solutions have a stronger interaction with the precipitant. Deterioration of precipitant quality during mixed solution precipitation is likely to cause a stall in cellulose phase separation. Based on this, we hypothesized that the introduction of additives of the synthetic polymer PAN into cellulose will facilitate the passage of competing processes during the separation of polymer phases and reduce the likelihood of defects (vacuoles) in the formed material when using a “rigid” precipitant (water). Thus, the goal of this research was to use optical interferometry to investigate the precipitation of mixed solutions with varied component ratios and compare the results to data from cellulose and PAN solutions in NMMO.

2. Materials and Methods

To prepare mixed solutions based on PAN and cellulose, powdered cellulose (Baikal Pulp and Paper Mill, Baykalsk, Russia) (degree of polymerization 600, moisture content ~8%, α-cellulose ~92%) and terpolymer PAN (Mw = 85,000 g/mol) (Good Fellow, Huntingdon, UK) were used. N-methylmorpholine-N-oxide (NMMO) was used as a solvent; T_m = 110–120 °C (water content ~10%), manufactured by Demochem (Shanghai, China). To inhibit the processes of thermo-oxidative destruction of cellulose, 0.5% propyl gallate (Sigma-Aldrich, St. Louis, MI, USA) was introduced into the system.

12% polymer solutions in a solvent with PAN/Cellulose component ratios of 0/100, 30/70, 50/50, 70/30, and 100/0 were produced using the method described in [37]. Distilled water was employed as a precipitant to form films.

The rheological behavior of the mixed solutions was studied using a Physica MCR 301 rotational rheometer (Anton Paar, Graz, Austria) (cone diameter of 25 mm and angle of 1°). The measurements were carried out in the shear deformation mode at a controlled shear stress in the range of 10–10³ Pa, as well as in the oscillation mode in the range of linear viscoelasticity (constant deformation amplitude of 1%, frequency range of 6–600 rad/s). The tests were carried out at a constant temperature of 110 °C. The rheometer was calibrated in accordance with existing metrological requirements using standardized control samples. The deviation in viscosity values obtained for the reference samples did not exceed 2%.

The process of precipitation of mixed solutions was modeled in the cell of an optical interferometer using the “drop” method described earlier [38]. The interaction of the system components was recorded by changing the bends of the interference fringes in the contact zone. The experiment was conducted at a temperature of 90 °C. The temperature of the precipitator was 25 °C.

3. Results

Cellulose and PAN, having different natures, dissolve in NMMO to form isotropic solutions. The morphological image of such cellulose solutions does not vary over time; but, for PAN solutions, prolonged exposure to high temperatures might cause a color shift. Therefore, for all solutions, the morphology was assessed immediately after their production (Figure 1).

Unlike PAN and cellulose solutions, mixed systems are emulsions. Emulsions’ morphology is determined by their solution preparation history. Intensive deformation of mixed solutions, as in a twin-screw extruder, provides for fine-droplet morphology, making it difficult to distinguish the matrix solution. This is evident from the photos of mixed solutions with cellulose-to-PAN ratio of 70/30 and 30/70. The average diameter of the dispersed phase droplets does not exceed 1 μm. For equiconcentrated solutions, the emulsion is an interpenetrating structure. In this concentration region, polymer phase inversion is common, as is a change in the system’s physical characteristics. As the concentration of PAN in the solution increases, droplet shape develops. Cellulose solution droplets are labile, allowing them to agglomerate and develop fibrillar morphologies when sheared.

The observed morphology of the solutions provides a deeper understanding of the rheological behavior of the solutions. Figure 2 shows the dependences of viscosity on shear stress for PAN, cellulose, and their mixed solutions in NMMO.

Under the current experimental conditions, PAN and cellulose solutions display nearly Newtonian behavior, with viscosity remaining constant throughout a large range of applied shear stresses. Mixed solutions of PAN/cellulose are characterized by a deviation from Newtonian behavior. Mixed solutions are microemulsions, which explains this behavior. In the region of high shear rates, such systems often exhibit the phenomenon of “slippage” along the interphase boundary and, as a consequence, a continuous decrease in viscosity with an increase in the shear rate [10]. Thus, the type of matrix polymer determines the nature of the curve. With an increase in shear stress (rate), the greater decrease in viscosity is observed the higher the cellulose content in the mixed solution. At the same time, for mixed solutions, destruction of the structure is observed at high stress (conventionally, “phase separation”).

Based on the results of rheological measurements of the mixtures, a concentration dependence was plotted reflecting the influence of the PAN/cellulose component ratio in the NMMO solution on the viscosity (Figure 3). The concentration dependence is common for systems with low- and high-viscosity components [39]: when the fraction of the low-viscosity component increases, the mixture’s viscosity declines.

It is interesting that the concentration dependence deviates from the additivity line, which connects the points corresponding to the viscosity of cellulose and PAN solutions in NMMO. The deviation from the additivity line increases with the increase in the proportion of cellulose in the system. Such a deviation of viscosity values is probably due to the interaction between the co-components of the solution.

Dynamic tests of the mixtures showed that the addition of a low-viscosity component (PAN) reduces the elasticity of the system (Figure 4).

Thus, for the cellulose solution, predominantly viscous behavior (G″ > G′) of the system is observed in the low-frequency region. At frequencies above 200 rad/s, the elastic component exceeds the viscous one. For mixed solutions, the crossover (G″ = G′) is recorded only for the system with a cellulose content of 70%. For other systems, the loss modulus exceeds the storage modulus in the entire frequency region. The predominantly viscous behavior of the systems can be traced in the dependences of the tangent of the angle of mechanical losses (Figure 5). At low frequencies, tgα takes values above 1.

The features of the rheological behavior of mixed solutions (the dependence of viscosity on the proportion of the dispersed phase and the rate of deformation) influence the precipitation processes of such systems.

According to the work [40], microinterferometry can be used for the mathematical description of the obtained interferograms for binary systems. In the case of ternary systems (pseudobinary), the use of interferometry is allowed in the case of compatibility of the components [41]. Modeling the precipitation process using the drop method made it possible to evaluate the evolution of diffusion processes occurring in polymer solutions upon contact with a “rigid” precipitant (water). Typical interferograms of the precipitation process for each of the solutions are presented in Figure 6, Figure 7 and Figure 8.

During the spinning process, the heated solution is passed through the channels or slit of the die and enters the precipitator through a small air gap. When passing through the air gap, the solution does not have time to cool down and, entering the bath, contacts the precipitator with a lower temperature. A similar picture is observed for all the systems under study. In the first seconds of contact between the components (solution/precipitant), active interdiffusion of the solvent and precipitant is recorded. This is displayed on the interferograms as a curvature of the interference bands on both sides of the phase boundary. In this case, a polymer film is formed in the diffusion zone due to the phase decomposition of the solution. Over time, an increase in the diffusion zone is recorded due to the interaction of the precipitant with the solvent and active growth of the polymer film. During the precipitation of cellulose solutions, vacuoles are formed, which is typical for rigid precipitation. The vacuoles begin to grow at a certain distance from the phase boundary. The increase in the size of the vacuoles is uneven, and the predominant direction of growth is close to perpendicular to the observed boundary. The height of the vacuoles one minute after the contact of the precipitant with the solution reaches 100 µm, and the thickness does not exceed 30 µm.

The obtained interferograms can be used to construct concentration profiles in the diffusion zone and to evaluate the kinetics of the process. However, such work has already been carried out, and the numerical characteristics of the process are described in [41,42]. Therefore, we will further focus on describing the evolution of the precipitation process and evaluating the propagation velocity of the precipitation fronts.

Figure 9 shows the morphology of a cross-section of a cellulose film obtained by precipitation in water.

The resulting cellulose film has asymmetry in height and contains at least 4 layers. The upper dense layer is formed during the first contact with the precipitant. Next, a layer consisting of vacuoles with a height of about 10–12 μm is observed. Below are layers with different densities but uniform texture.

In [36] it was shown that PAN interacts more actively with NMMO, and less precipitant is required to initiate its coagulation. The interferograms (Figure 7) show the evolution of precipitation of 12% PAN solutions in NMMO.

The viscosity of 12% PAN solutions is significantly lower than that of cellulose solutions, so the precipitant must diffuse at a higher rate into the solution volume. Already, 10 s after the first contact between the precipitant and the solution, the height of the vacuoles reaches 250 μm. Comparing the height of the vacuoles for cellulose- and PAN-based systems, we can talk about their active formation in low-viscosity PAN solutions in NMMO. Further precipitation of the solution is accompanied by the growth of vacuoles, i.e., a more defective morphology is formed.

If for cellulose and PAN solutions in NMMO, defective (permeated with vacuoles) structures are formed (Figure 6 and Figure 7), then for mixed systems, on the contrary, a dense monolithic morphology is observed (Figure 8, Figure 10 and Figure 11).

Mixed solutions with PAN content of 30% retain rheological properties typical for cellulose solutions, i.e., high viscosity and elasticity compared to PAN solutions. On the other hand, a low-viscosity solution of PAN in NMMO will interact with water much more actively. Consequently, the precipitation process will proceed by a different mechanism compared to cellulose (PAN) solutions in NMMO. The cellulose phase will be released after PAN precipitation upon contact with an aqueous solution of NMMO formed after PAN precipitation. It is known that a softer precipitant leads to the formation of a more uniform morphology, which is observed in the interferograms.

With an increase in the proportion of PAN in the system for equiconcentrated systems, the precipitation of the cellulose phase will occur under milder conditions (Figure 10).

The viscosity of a mixed solution containing 50% PAN is about 20 Pa*s at 110 °C, which is several times less than that of cellulose solutions. For such a system, one can speak of interpenetrating morphology, where it is not possible to determine the dispersed phase. A large proportion of the PAN solution will promote the formation of a more concentrated aqueous solution of NMMO during precipitation, with which the cellulose solution subsequently interacts. An increase in the NMMO content in the precipitant promotes an even softer precipitation of the cellulose phase.

Increasing the PAN content in the system to 70% allows to obtain an emulsion, where the cellulose solution in NMMO becomes a dispersed medium (Figure 11).

The viscosity of such solutions approaches the values for 12% PAN solutions in NMMO, while the elastic modulus is an order of magnitude higher. Cellulose, being in the volume of the PAN solution, will precipitate only after the PAN phase. An aqueous NMMO solution, contacting with drops of cellulose solution, will precipitate them at a lower rate.

It has already been mentioned above that the microinterferometry method is usually used for binary systems or ternary systems in case of component compatibility. For systems of 4 components, which contain two polymers, this method has not been used before. It is difficult to apply known mathematical descriptions in such a case. However, the observed picture allows us to obtain quantitative information about the process and observe the morphology being formed in the system.

The composition of the polymer mixture determines the rate of diffusion processes and, as a consequence, the rate of polymer “precipitation”. The kinetics of the precipitated film growth, as well as the rate of formation of the precipitation zone, are shown in Figure 12 and Figure 13, respectively. The highest linear rate of film formation is observed for the PAN solution. The evolution of the precipitation front position allows us to observe the influence of the dispersed phase in the mixed emulsion. Notwithstanding the intricacy of the process, wherein alterations in the percentage of the dispersed phase affect both the rheological properties of the solutions and the competition among the released phases for the precipitant, one may discuss their impact on the resultant morphology.

The translational mobility of the components in this case is limited by the viscosity of the solution. The lowest viscosity, according to rheological measurements, is recorded for the PAN solution. The diffusion process is most intense in the first minutes of contact, then a slowdown in diffusion processes is observed (Figure 14), which is associated with the gradual formation of a polymer gel zone. Despite the different nature of the polymers, the highest and lowest values are observed for cellulose solutions and systems with 50% PAN content. This is probably due to the step-by-step separation of polymer phases. First, a PAN solution with a viscosity lower than for a 12% PAN solution in NMMO is precipitated. Then, coagulation of the cellulose phase occurs.

When comparing the precipitation rates (growth of the zone of the precipitated system), it is clear that its values are influenced by the proportion of PAN and its distribution in the volume of the mixed solution.

The kinetics of the movement of isoconcentration planes in the coordinates of the diffusion equation Δx~t^1/2 (where Δx is the depth of penetration of the diffusing component, t is the observation time) is linear (Figure 14) [43]. The speed of movement of diffusion flows is determined by the individual characteristics of each of the systems.

Thus, using the optical interferometry method, it is possible not only to trace the precipitation kinetics for cellulose, PAN and mixed solutions in NMMO, but also to observe the suppression of defect formation in the precipitated system. The introduction of a soluble additive affects the viscoelastic properties of the mixed solution and the rate of its precipitation. For equiconcentrated solutions, the rate of polymer precipitation is the lowest, which is explained by the synergism of the rheological properties of the system and the active interaction of PAN solutions in NMMO with water.

4. Conclusions

For cellulose and PAN solutions, precipitation processes are well known. When precipitating 12% solutions with a rigid precipitant (water), vacuole defects are formed, and the morphology of the formed film is asymmetric. Changing the polarity of nonsolvents (precipitants) or their temperature allows reducing the mutual diffusion of the solvent and precipitant in the system. For mixed solutions based on PAN and cellulose, the separation of polymer phases using precipitants is a non-trivial issue requiring detailed study. It has been shown for the first time that the precipitation of emulsions based on cellulose and PAN with water is affected by the ratio of its components. With an increase in the proportion of PAN in mixed solutions, the viscosity and elasticity of the mixed systems initially decrease insignificantly with a significant decrease in the viscosity of the dispersed medium. A further increase in the proportion of PAN leads to a drop in the viscosity of the emulsion and the solutions forming it. Reducing the viscosity of the emulsion components accelerates the precipitation of PAN and cellulose solutions in NMMO. The rate of PAN release from solutions is higher than that of cellulose, and the amount of water required to start the process is less. As a result, cellulose in emulsions with a PAN share of 50% or more will probably be precipitated not by water, but by an aqueous solution of NMMO (a softer precipitant). Such precipitation will lead to non-uniform precipitation of the PAN and cellulose phases, and the properties of the formed material are less predictable. When the PAN content in the system is 30% or more, the formation of defects in the precipitated solution is significantly suppressed, which is crucial for achieving a uniform morphology. In the future, targeted selection of PAN copolymers and nonsolvent compositions (their temperatures) will allow the formation of the required structure of fiber, film or membrane.